Publication number | US20030222807 A1 |

Publication type | Application |

Application number | US 10/159,444 |

Publication date | Dec 4, 2003 |

Filing date | May 31, 2002 |

Priority date | May 31, 2002 |

Also published as | DE60315215D1, DE60315215T2, EP1367409A2, EP1367409A3, EP1367409B1, US6661369 |

Publication number | 10159444, 159444, US 2003/0222807 A1, US 2003/222807 A1, US 20030222807 A1, US 20030222807A1, US 2003222807 A1, US 2003222807A1, US-A1-20030222807, US-A1-2003222807, US2003/0222807A1, US2003/222807A1, US20030222807 A1, US20030222807A1, US2003222807 A1, US2003222807A1 |

Inventors | Kwang Cho |

Original Assignee | Cho Kwang M. |

Export Citation | BiBTeX, EndNote, RefMan |

Referenced by (7), Classifications (9), Legal Events (4) | |

External Links: USPTO, USPTO Assignment, Espacenet | |

US 20030222807 A1

Abstract

A system and method (**44**) for focusing an image oriented in an arbitrary direction when the collected synthetic aperture radar (SAR) data is processed using range migration algorithm (RMA). In accordance with the teachings of the present invention, first (**60**) the data is skewed so that the direction of smearing in the image is aligned with one of the spatial frequency axes of the image. In the illustrative embodiment, the smearing is aligned in the vertical direction. This is done through a phase adjustment that was derived from the requirements for proper shift in the spatial frequency domain. Next (**62**), the signal support areas from all targets are aligned by proper phase adjustment in the spatial (or image) domain. Finally (**64**), the common phase error can be corrected using autofocus algorithms.

Claims(24)

first means for skewing said image;

second means for aligning said signal support areas from all targets after said image has been skewed; and

third means for correcting common phase error after said signal support areas have been aligned.

fourth means for converting input data to the (X, K_{Y}) domain, and

fifth means for applying a phase adjustment to said converted data.

sixth means for converting skewed data to the spatial domain (X, Y), and

seventh means for applying a phase shift to the resulting spatial domain data.

where λ is the wavelength of the transmitted radar signal and R is the distance from the radar platform to the image center.

eighth means for converting phase shifted data to the (K_{X}, Y) domain, and

ninth means for applying an autofocus function to said converted data.

eleventh means for converting phase corrected data to the spatial domain (X,Y), and

twelfth means for shifting the phase of the resulting spatial domain data.

fourteenth means for converting phase corrected data to the (X,K_{Y}) domain, and

fifteenth means for applying a reverse phase adjustment to the converted data.

means for applying weights and an inverse Fast Fourier Transform in K_{Y}, generating data in the image (X, Y) domain;

means for converting said image domain data to the (K_{X}, Y) domain; and means for applying weights and an inverse Fast Fourier Transform in K_{X}, generating a final focused image.

means for converting said phase corrected data to image (X, Y) domain, and

means for applying a two dimensional Spatially Variant Apodization function, generating a final focused image.

an antenna for transmitting and receiving electromagnetic energy;

a transmitter for sending transmit information to said antenna;

a receiver for processing said received electromagnetic energy; and

a signal processor with a range migration algorithm containing a phase correction system adapted to execute:

a first function for skewing the image data;

a second function for aligning the signal support areas from all targets after the data has been skewed; and

a third function for correcting common phase error after said signal support areas have been aligned.

skewing said image;

aligning said signal support areas from all targets after said image has been skewed; and

correcting common phase error after said signal support areas have been aligned.

Description

- [0001]1. Field of the Invention
- [0002]The present invention relates to radar signal processing systems. More specifically, the present invention relates to phase error correction in a range migration algorithm (RMA) for synthetic aperture radar (SAR) systems.
- [0003]2. Description of the Related Art
- [0004]In imaging applications such as ground mapping, a radar system is used to generate a two-dimensional image of a portion of a ground surface in the range and azimuth (cross-range) directions. A large antenna aperture is required in conventional imaging radar systems in order to achieve a narrow beamwidth and, consequently, fine azimuth resolution. Synthetic aperture radar (SAR) systems have been developed as an alternative means for improving azimuth resolution by synthesizing pulse-to-pulse return signals collected by a moving platform with a small antenna. Signal synthesis from many successive locations of the moving platform accomplishes what would otherwise require a larger antenna aperture.
- [0005]A polar format algorithm (PFA) has been widely used for high-resolution SAR systems. However, the PFA has the shortcomings of limited depth of focus and geometric distortion that increases with the map size. Range migration algorithm (RMA) is one of the most attractive and advanced SAR processing techniques to avoid the problems with the PFA. A difficulty with the RMA, however, is performing efficient phase correction.
- [0006]Normal SAR data collection requires phase coherence, not only within each pulse for range resolution, but also from pulse to pulse over the collection time needed for azimuth resolution. The platform position affects the pulse-to-pulse phase coherence over the synthetic aperture. Phase error introduced by the inaccuracy of navigation data or undesirable platform motion causes smearing or duplication of the target image. Since motion compensation at the early processing stage based on the navigation data is not sufficient for producing a focused image, it is a common practice to employ data driven autofocus algorithms in high resolution SAR systems in order to maintain phase coherence and achieve good image quality.
- [0007]Considering the computational efficiency and simplicity, it is desirable to implement the autofocus function after range compression during batch processing as is usually done for the case of PFA. However, for the case of RMA, it is difficult to implement the autofocus function during batch processing because the signal support areas from different targets are not aligned. For this reason, the autofocus function had to be implemented before batch processing starts, at the cost of increased complexity and processing time.
- [0008]Because of the difficulty in implementing the autofocus function during batch processing, current RMA systems perform the autofocus function during the pulse-to-pulse processing phase using a separate polar format processing algorithm. This approach, however, has the disadvantage of implementation complexity and tighter processing timeline requirements. Furthermore, the increased timeline requirement makes it more difficult, if not impossible, to implement more advanced autofocus techniques.
- [0009]In a patent application entitled EFFICIENT PHASE CORRECTION SCHEME FOR RANGE MIGRATION ALGORITHM, application Ser. No. 10/060,647, filed Jan. 30, 2002, by K. M. Cho (Atty. Docket No. PD 01W053), the teachings of which are incorporated herein by reference, an efficient phase error correction scheme for RMA is disclosed which allows phase correction to be performed during batch processing. This approach requires the image to be oriented in range-azimuth. However, for current and proposed applications, there is a need for images to be formed in a direction other than range and azimuth.
- [0010]In the general case when the processed image is oriented in an arbitrary direction especially with RMA, phase correction is more complicated and requires additional processing. A currently used method performs phase correction through separate processing in the pulse-to-pulse phase using PFA. There is no known method for phase correction of arbitrarily oriented images in RMA during batch processing.
- [0011]Hence, a need exists in the art for an improved system or method for efficient phase error correction of arbitrarily oriented images in a range migration algorithm.
- [0012]The need in the art is addressed by the present invention, a system and method for focusing an image oriented in an arbitrary direction when the collected synthetic aperture radar (SAR) data is processed using range migration algorithm (RMA). In accordance with the teachings of the present invention, first the data is skewed so that the direction of smearing in the image is aligned with one of the spatial frequency axes of the image. In the illustrative embodiment, the smearing is aligned in the vertical direction. This is done through a phase adjustment that was derived from the requirements for proper shift in the spatial frequency domain. Next, the signal support areas from all targets are aligned by proper phase adjustment in the spatial (or image) domain. The common phase error is then estimated and corrected using autofocus algorithms. The remaining steps include reverse shifting and deskew processing for the reversal of the processing performed earlier for the alignment of common phase error.
- [0013][0013]FIGS. 1
*a*and**1***b*are illustrations showing the direction of smearing in an image I oriented in range-azimuth. - [0014][0014]FIGS. 1
*c*and**1***d*are illustrations showing the direction of smearing in an image I′ oriented in along track-cross track. - [0015][0015]FIG. 2 is an illustration showing a general case that has a Doppler cone angle φ and an image orientation angle φ
_{1 }from the along-track direction V. - [0016][0016]FIG. 3 is an illustration showing the direction of pulse-to-pulse phase variation in the spatial frequency domain.
- [0017][0017]FIG. 4 illustrates the data conversion from (K
_{X}, K_{Y}) to ({circumflex over (K)}_{X}, {circumflex over (K)}_{Y}) to align the phase error variation to the vertical direction. - [0018][0018]FIG. 5 shows the geometry for an image with arbitrary orientation with targets a and b at the map center (0,0) and at (X,Y), respectively.
- [0019][0019]FIG. 6 is an illustration showing the signal support areas A and B corresponding to targets a and b, respectively.
- [0020][0020]FIG. 7 is an illustration showing the shifting of the signal support in K
_{X }only as a function of target location in cross range. - [0021][0021]FIG. 8 is a simplified block diagram of an airborne SAR system using RMA with the new phase correction scheme in accordance with the teachings of the present invention.
- [0022][0022]FIG. 9 is a simplified block diagram showing the basic components of RMA and where the new phase correction scheme should be inserted in accordance with the teachings of the present invention.
- [0023][0023]FIG. 10 is a simplified block diagram showing the basic components of the new phase correction scheme in accordance with the teachings of the present invention.
- [0024][0024]FIGS. 11
*a*-**11***c*show the data length variation for the case with minimum spectral data size. - [0025][0025]FIG. 11
*a*shows the interpolated data after the Stolt interpolation in the (K_{X}, K_{Y}) domain. - [0026][0026]FIG. 11
*b*shows the data position after data skewing to change smearing to the vertical direction. - [0027][0027]FIG. 11
*c*shows the data position after data shifting in K_{X}. - [0028][0028]FIGS. 12
*a*-**12***c*show the data length variation for the case with extended spectral data size. - [0029][0029]FIG. 12
*a*shows the interpolated data after the Stolt interpolation in the (K_{X}, K_{Y}) domain. - [0030][0030]FIG. 12
*b*shows the data position after data skewing to change smearing to the vertical direction. - [0031][0031]FIG. 12
*c*shows the data position after data shifting in K_{X}. - [0032][0032]FIGS. 13
*a*and**13***b*are flow diagrams of the new phase correction algorithm in accordance with the teachings of the present invention. The flow chart begins in FIG. 13*a*and is continued in FIG. 13*b.* - [0033][0033]FIG. 14 is a flow chart of a particular embodiment of the present invention showing the simplified processing sequence for focusing RMA-based SAR images oriented in range-azimuth.
- [0034]Illustrative embodiments and exemplary applications will now be described with reference to the accompanying drawings to disclose the advantageous teachings of the present invention.
- [0035]While the present invention is described herein with reference to illustrative embodiments for particular applications, it should be understood that the invention is not limited thereto. Those having ordinary skill in the art and access to the teachings provided herein will recognize additional modifications, applications, and embodiments within the scope thereof and additional fields in which the present invention would be of significant utility.
- [0036]Uncompensated phase error in collected data for synthetic aperture radar (SAR) systems causes image smearing in the azimuth (cross-range) direction. This is due mainly to uncompensated slow motion error that occurs as the Doppler cone angle changes during the time of data collection. When SAR images are formed in the range and azimuth direction, image smearing that occurs in the azimuth direction (usually the vertical axis in the SAR display) can readily be corrected by compensating estimated phase error in the azimuth direction only using various autofocus techniques. However, when the formed image is oriented in a direction different from range and azimuth, the azimuth response that reflects pulse-to-pulse phase variation is not aligned to the vertical axis and image smearing cannot be corrected simply by one-dimensional processing. The direction of smearing is illustrated in FIG. 1 for images oriented in range-azimuth and along track-cross track.
- [0037][0037]FIGS. 1
*a*and**1***b*are illustrations showing the direction of smearing in an image I oriented in range-azimuth with a Doppler cone angle φ measured from the along track direction V. The direction of smearing is in the azimuth direction, which is along the vertical axis of the image as shown in FIG. 1*b.* - [0038][0038]FIGS. 1
*c*and**1***d*are illustrations showing the direction of smearing in an image I′ oriented in along track-cross track (AT-CT) with a Doppler cone angle φ measured from the along track direction V. The direction of smearing is still in the azimuth direction. However, since the image is not oriented in range-azimuth, the direction of smearing is not along the vertical axis of the image as shown in FIG. 1*d.* - [0039]Since the direction of image smearing does not align to that of the image pixels when the images are oriented in a direction other than range and azimuth, direct application of autofocus algorithms in the azimuth direction as conventionally done does not lead to resolving the smearing problem. The proposed method performs a data skew by applying proper phase adjustments so that the direction of smearing is converted to the desirable vertical direction before applying autofocus algorithms. Inverse processing needs to be performed after phase correction using autofocus algorithms. When the range migration algorithm (RMA) is employed for image formation, the common phase errors in the phase history of each target are not aligned because of the shifting of the signal support in the spatial frequency domain that is determined by the azimuth location of the target. Therefore, to estimate and correct the uncompensated common phase error, it is required to shift the signal support for each target so that the common phase error is aligned. Shifting of the signal support for the RMA processing includes basic shifting of signal support and secondary shifting due to the conversion of the smearing direction as mentioned.
- [0040][0040]FIG. 2 is an illustration showing a general case that has a Doppler cone angle φ and an image orientation angle φ
_{1 }from the along-track direction V. The Y axis of the image is at an angle φ_{1 }from the along-track direction V, and the X axis of the image is at an angle φ_{1}−φ from the azimuth direction. Since the spectrum corresponding to the processed image in the spatial frequency domain before the last two-dimensional inverse FFT in the image formation processing chain has the same orientation with the image orientation, pulse-to-pulse phase error occurs at an angle φ_{1}−φ from the vertical axis (K_{X}) as indicated in FIG. 3. - [0041][0041]FIG. 3 is an illustration showing the direction of pulse-to-pulse phase variation in the spatial frequency domain.
- [0042]Since the pulse-to-pulse phase variation occurs in a direction that couples two axes, the image cannot be focused by one-dimensional phase correction. Therefore, to use one-dimensional phase correction algorithms, it is required to convert the data such that the direction of phase variation can be changed to either the vertical or horizontal direction.
- [0043][0043]FIG. 4 illustrates the data conversion from (K
_{X}, K_{Y}) to ({circumflex over (K)}_{X}, {circumflex over (K)}_{Y}) to align the phase error variation to the vertical direction. To make the phase vary in the vertical direction, the following substitution is required: -
*K*_{X}*={circumflex over (K)}*_{X}−tan(φ_{1}−φ)·*K*_{Y}(1) - [0044]Although not presented here, this requirement can also be obtained analytically by considering error effect when there is an AT velocity error.
- [0045]Eq. (1) requires K
_{X }shifted as a function of K_{Y }and it makes the phase {circumflex over (θ)} changed to:$\begin{array}{cc}\hat{\theta}=-2\ue89e\pi \ue8a0\left({\hat{K}}_{X}\xb7X+{K}_{Y}\xb7Y\right)=-2\ue89e\pi \ue8a0\left[{K}_{X}\xb7X+{K}_{Y}\xb7\left(Y+\mathrm{tan}\ue8a0\left({\varphi}_{1}-\varphi \right)\xb7X\right)\right]& \left(2\right)\end{array}$ - [0046]Therefore, the required phase change from θ=−2π(K
_{X}X+K_{Y}Y) to {circumflex over (θ)}=θ+Δθ is: - Δθ=−2π tan(φ
_{1}−φ)*K*_{Y}*X*(3) - [0047]The expression for {circumflex over (θ)} also indicates that the added linear phase makes the image shifted in the horizontal (Y) direction as a function of vertical (X) position of targets. Calling this data conversion process by skew process, deskew processing is required after estimating and eliminating phase error that is common to all targets using any autofocus algorithms.
- [0048]Since supporting spectra corresponding to targets at different azimuth locations are shifted from each other, it is required to perform the proper shift so that phase history from different targets are aligned for the estimation and correction of the common phase error.
- [0049][0049]FIG. 5 shows the geometry for an image with arbitrary orientation with targets a and b at the map center (0,0) and at coordinates (X,Y), respectively. Target a is located at an angle φ from the along-track direction V, at a distance R from the radar. Target b is located at an angle φ
_{1 }from the along-track direction V. The mapping geometry in FIG. 5 will be used to derive the required shift correction based on the calculated target angle off line-of-sight (LOS). - [0050]From the expressions for the spatial frequency variables, K
_{X}=−(2/λ)cos φ and K_{Y}=(2/λ)sin φ, the signal support for targets at the map center and at (X,Y) can be illustrated as shown in FIG. 6 when the range separation is not that large compared with the mapping range. - [0051][0051]FIG. 6 is an illustration showing the signal support areas A and B corresponding to targets a and b, respectively. The signal support areas A and B are shifted by an angle φ−φ
_{1}. The shifting of the signal support is determined by the azimuth location of the targets. - [0052]Since the phase error was changed to align to the vertical (K
_{X}) direction as discussed above, shifting of the signal support can be done in the vertical direction. - [0053]
- [0054]It should be noted that the spatial frequency in the cross-range direction K
_{AZ }varies in the direction of angular change and its shifting for a target off LOS by Δφ is expressed as ΔK_{AZ}=2Δφ/λ. Then, the shifting in K_{X }is given by:$\begin{array}{cc}\Delta \ue89e\text{\hspace{1em}}\ue89e{K}_{X}=\mathrm{cos}\ue8a0\left({\varphi}_{1}-\varphi \right)\xb7\Delta \ue89e\text{\hspace{1em}}\ue89e{K}_{\mathrm{AZ}}=\frac{2\ue89e\text{\hspace{1em}}\ue89e\mathrm{cos}\ue8a0\left({\varphi}_{1}-\varphi \right)\ue89e\mathrm{\Delta \varphi}}{\lambda}& \left(5\right)\end{array}$ - [0055]Therefore, the phase adjustment required in the spatial domain to compensate for a shift in K
_{X }is given by:$\begin{array}{cc}\begin{array}{c}{\mathrm{\Delta \Phi}}_{1}=2\ue89e\pi \xb7{K}_{X}\ue89eX\\ =\frac{4\ue89e\mathrm{\pi cos}\ue8a0\left({\varphi}_{1}-\varphi \right)\xb7\mathrm{\Delta \varphi}\xb7X}{\lambda}\\ =\frac{4\ue89e\mathrm{\pi cos}\ue8a0\left({\varphi}_{1}-\varphi \right)}{R\ue89e\text{\hspace{1em}}\ue89e\lambda}\ue8a0\left[\mathrm{sin}\ue8a0\left({\varphi}_{1}-\varphi \right)\xb7X\xb7Y-\mathrm{cos}\ue8a0\left({\varphi}_{1}-\varphi \right)\xb7{X}^{2}\right]\end{array}& \left(6\right)\end{array}$ - [0056]
- [0057]Then ΔΦ
_{1 }has the expression:$\begin{array}{cc}{\mathrm{\Delta \Phi}}_{1}=\frac{4\ue89e\pi \ue89e\text{\hspace{1em}}\ue89e\mathrm{cos}\ue89e\text{\hspace{1em}}\ue89e\left({\varphi}_{1}-\varphi \right)}{R\ue89e\text{\hspace{1em}}\ue89e\lambda}\ue8a0\left[\mathrm{sin}\ue8a0\left({\varphi}_{1}-\varphi \right)\xb7X\xb7Y-\frac{1}{2}\ue89e\mathrm{cos}\ue8a0\left({\varphi}_{1}-\varphi \right)\xb7{X}^{2}\right]& \left(7\right)\end{array}$ - [0058]Application of this phase adjustment makes the signal support in the spatial frequency domain for targets located at different azimuth positions aligned as illustrated in FIG. 7. FIG. 7 is an illustration showing the shifting of the signal support in K
_{X }only as a function of target location in cross range. - [0059]An additional phase adjustment is required for ΔK
_{X}=tan(φ_{1}−φ)ΔK_{Y }in (1) using ΔK_{Y}=2 sin(φ_{1}−φ)Δφ/λ:$\begin{array}{cc}\begin{array}{c}{\mathrm{\Delta \Phi}}_{2}=2\ue89e\pi \xb7\Delta \ue89e\text{\hspace{1em}}\ue89e{K}_{X}\ue89eX\\ =2\ue89e\mathrm{\pi tan}\ue8a0\left({\varphi}_{1}-\varphi \right)\ue89e\Delta \ue89e\text{\hspace{1em}}\ue89e{K}_{Y}\ue89eX\\ =\frac{4\ue89e\mathrm{\pi tan}\ue8a0\left({\varphi}_{1}-\varphi \right)\ue89e\mathrm{sin}\ue8a0\left({\varphi}_{1}-\varphi \right)}{R\ue89e\text{\hspace{1em}}\ue89e\lambda}\ue8a0\left[\mathrm{sin}\ue8a0\left({\varphi}_{1}-\varphi \right)\xb7X\xb7Y-\mathrm{cos}\ue8a0\left({\varphi}_{1}-\varphi \right)\xb7{X}^{2}\right]\end{array}& \left(8\right)\end{array}$ - [0060]Combining the expressions in (7) and (8), the following result is obtained for total phase adjustments:
$\begin{array}{cc}\mathrm{\Delta \Phi}={\mathrm{\Delta \Phi}}_{1}+{\mathrm{\Delta \Phi}}_{2}=\frac{4\ue89e\pi}{R\ue89e\text{\hspace{1em}}\ue89e\lambda}\ue89e\hspace{1em}\left[X\xb7Y\xb7\mathrm{tan}\ue8a0\left({\varphi}_{1}-\varphi \right)-{X}^{2}\ue89e{\mathrm{cos}}^{2}\ue8a0\left({\varphi}_{1}-\varphi \right)\ue89e\left(\frac{1}{2}+{\mathrm{tan}}^{2}\ue8a0\left({\varphi}_{1}-\varphi \right)\right)\right]& \left(9\right)\end{array}$ - [0061]Applying this phase function in the spatial domain will make the shifted spectra aligned so that the common phase error can be estimated and compensated.
- [0062]The present invention provides a system and method for efficient phase error correction in RMA for images with arbitrary orientation. First, the proposed method performs data skew by applying the proper phase adjustment, as discussed above, so that the direction of smearing is converted to the desirable vertical direction. Second, the signal support for each target is shifted so that the common phase error is aligned. The shifting of the signal support includes both basic shifting of signal support and secondary shifting due to the conversion of the smearing direction as discussed above. Then, the common phase error can be corrected using known autofocus algorithms. Reverse processing of the previously performed phase adjustments needs to be performed after phase correction.
- [0063][0063]FIG. 8 is a simplified block diagram of an airborne SAR system using RMA with the new phase correction scheme in accordance with the teachings of the present invention. A SAR system
**16**mounted on an aircraft**10**generates an electromagnetic wave**12**which is reflected off a ground surface**14**and received by the SAR system**16**. The SAR system**16**includes an antenna**20**, a transmitter**22**, and a receiver**24**for radiating and receiving the electromagnetic waves. The receiver**24**forms input video phase history (VPH) data**28**from the received data and sends it to an on-board signal processor**26**. Residing in the signal processor**26**is a range migration algorithm**30**with the new phase correction scheme of the present teachings, which outputs a focused image**50**. - [0064][0064]FIG. 9 is a simplified block diagram showing the basic components of RMA and where the new phase correction scheme should be inserted in accordance with the teachings of the present invention. The basic components of RMA are well known in the art. The input VPH data
**28**is successively passed through a Motion Compensation function**32**, a Range Deskew function**34**, an Along-Track FFT**26**, a Matched Filter**38**, and a Stolt Interpolation**40**. The new phase correction scheme**44**is inserted after the Stolt Interpolation**40**, generating a final focused image**50**. - [0065][0065]FIG. 10 is a simplified block diagram showing the basic components of the new phase correction scheme
**44**in accordance with the teachings of the present invention. Interpolated data**58**output from the Stolt Interpolation**40**is first passed through a Skew function**60**, followed by a Phase Shift**62**, and an Autofocus function**64**. After Autofocus**64**, a Reverse Shift**66**and Deskew**68**are applied. Finally, a 2D FFT**70**is applied to form a focused image**50**. - [0066]The proposed phase correction method needs to be implemented in the spatial frequency domain after the Stolt Interpolation
**40**. Proper zero padding is required to avoid the wrap-around effect in the process of performing FFT functions. In taking input data after the Stolt interpolation, there are two cases. The first case is when the interpolated data in the spatial frequency domain (K_{X}, K_{Y}) has the minimum size in K_{X }that is required for the desired resolution. Since phase error is estimated using the shifted spectra of each target, the estimated phase error in this case may be less accurate around the azimuth edge in the spatial frequency domain. The second case is when the interpolated data in the spatial frequency domain has extra length in K_{X }so that the overlapped area is big enough for phase correction of the desired length during and after forward and backward shifting. These two cases will be discussed first before presenting detailed functional steps. - [0067][0067]FIGS. 11
*a*-**11***c*show the data length variation for the case with minimum spectral data size. FIG. 11*a*shows the interpolated data after the Stolt interpolation in the (K_{X}, K_{Y}) domain. The data lengths in K_{X }and K_{Y }are DK_{Xmin }and DK_{Y}, respectively. - [0068][0068]FIG. 11
*b*shows the data position after data skewing to change smearing to the vertical direction. The data length in K_{X }becomes DK_{X min}+DK_{Y }tan(φ_{1}−φ). - [0069][0069]FIG. 11
*c*shows the data position after data shifting in K_{X}. The data length in K_{X }becomes DK_{X min}+DK_{Y }tan(φ_{1}−φ)+2(2/λ)ΔΦ_{max}. Because of the position dependent shifting, not all targets can provide phase information that is required for the estimation and correction of the common phase error in the initial spectral area. Therefore, in this case, there may be some degree of degradation in phase correction performance. However, this case offers a computational advantage when spatially variant apodization (SVA) is employed since one forward and one backward FFT functions in the last stage of processing in the other case are not required. - [0070]In the previous case, uncovered spectral regions due to target dependent shifting leads to degradation of phase correction performance. This can be avoided by taking spectral data with extended length.
- [0071][0071]FIGS. 12
*a*-**12***c*show the data length variation for the case with extended spectral data size. FIG. 12*a*shows the interpolated data after the Stolt interpolation in the (K_{X}, K_{Y}) domain. The data lengths in K_{X }and K_{Y }are DK_{X }and DK_{Y}, respectively, where DK_{X}=DK_{X min}+2(2/λ)ΔΦ_{max}. - [0072][0072]FIG. 12
*b*shows the data position after data skewing to change smearing to the vertical direction. The data length in K_{X }becomes DK_{X}+DK_{Y }tan(φ_{1}−φ). - [0073][0073]FIG. 12
*c*shows the data position after data shifting in K_{X}. The data length in K_{X }becomes DK_{X}+DK_{Y }tan(φ_{1}−φ)+2(2/λ)ΔΦ_{max}. Since phase information from every target in the mapping area is available after spectral shifting, phase correction performance should be better than the case with the minimum spectral size. There is still a small area that is not covered by the data collection as indicated in the first spectral rectangle in FIG. 12*a*. If desired, this can be avoided by taking extra samples in K_{Y }as indicated by dotted lines. - [0074][0074]FIGS. 13
*a*and**13***b*are flow diagrams of the new phase correction algorithm in accordance with the teachings of the present invention. The flow chart begins in FIG. 13*a*and is continued in FIG. 13*b.* - [0075]In STEP
**1**, data from the Stolt interpolation is taken as input. As discussed earlier, there are two cases with different sizes of spectra. The desired resolution δ_{X }determines the required minimum length DK_{X min }of the interpolated data in the spatial frequency domain: DK_{X min}=k_{X}/δ_{X }for the mainlobe broadening factor k_{X}. Taking more data than the required minimum allows more target phase information and therefore better phase estimation. Extended data length on each side can be calculated by the following expression derived from Eqn. (9):$\begin{array}{cc}\Delta \ue89e\text{\hspace{1em}}\ue89e{K}_{X}=\frac{2\ue89e\mathrm{\Delta \Phi}}{2\ue89e\pi \xb7X}\ue89e{|}_{X={W}_{X}/2,Y={W}_{Y}/2}\ue89e\hspace{1em}=\frac{2}{R\ue89e\text{\hspace{1em}}\ue89e\lambda}\ue8a0\left[{W}_{Y}\ue89e\mathrm{tan}\ue8a0\left({\varphi}_{1}-\varphi \right)-{W}_{X}\ue89e{\mathrm{cos}}^{2}\ue8a0\left({\varphi}_{1}-\varphi \right)\ue89e\left(\frac{1}{2}+{\mathrm{tan}}^{2}\ue8a0\left({\varphi}_{1}-\varphi \right)\right)\right]\ue89e\text{\hspace{1em}}& \left(10\right)\end{array}$ - [0076]Here W
_{X }and W_{Y }are the image size in each direction. Data length for the case of extended input can be expressed by DK_{X}=DK_{X min}+2ΔK_{X}. - [0077]In STEP
**2**, an inverse Fast Fourier Transform (IFFT) function in K_{X }is applied to the data to convert it to the spatial domain so that the phase adjustment Δθ in Eqn. (3) can be applied in STEP**3**for shifting in K_{X }as a function of K_{Y}. After this step, the data is in the (X, K_{Y}) domain. - [0078]In STEP
**3**, the phase adjustment Δθ in Eqn. (3) is applied. Application of this phase in the (X, K_{Y}) domain is for skewing the data to make the angled smearing aligned to the vertical direction. - [0079]In STEP
**4**, an inverse Fast Fourier Transform (IFFT) in K_{Y }is applied to convert the data to the image (X, Y) domain so that the phase adjustment in STEP**5**can be performed to align the shifted common phase error as a function of azimuth position. - [0080]In STEP
**5**, the phase adjustment function ΔΦ expressed in Eqn. (9) is applied to align the shifted target spectra so that the common phase error can be properly estimated and corrected using an autofocus function in STEP**7**. - [0081]In STEP
**6**, a forward FFT1 function in X is applied to convert the data in X to K_{X }so that the common phase error can be estimated and corrected by the subsequent autofocus function (STEP**7**). After this step, the data is in the (K_{X}, y) domain. - [0082]In STEP
**7**, any parametric or non-parametric autofocus algorithm or combination of algorithms can be applied to estimate and correct the phase error common to all targets using the shift-corrected range compressed data in the K_{X }domain. - [0083]In STEP
**8**, an inverse FFT in K_{X }is applied to convert the phase corrected data to the image (spatial) domain so that a reverse phase shift (−ΔΦ) can be applied in STEP**9**to restore the shifted signal support. - [0084]In STEP 9, the inverse function (−ΔΦ) of the phase adjustment performed earlier (STEP
**5**) is applied to shift back the signal support that is dependent on the target azimuth location. - [0085]In STEP
**10**, a forward FFT function in Y is applied to convert data from Y to K_{Y }so that the deskew function (−ΔΦ) in STEP**11**can be applied. After this step, the data is in the (X, K_{Y}) domain. - [0086]In STEP
**1**, a deskew phase function (−ΔΦ) is applied to shift back the shifted image in Y for a given position in X, reversing the skew function applied in STEP**3**. - [0087]In STEP
**12**, if SVA is used with the input spectrum of minimum length, go to STEP**16**; otherwise, go to STEP**13**. - [0088]In STEP
**13**, weights and an inverse FFT in K_{Y }are applied to compress the data in Y. After this step, the data is in the image domain (X, Y). - [0089]In STEP
**14**, a forward FFT in X is applied to convert the data to K_{X }so that data of the correct size is taken and a weighting function can be applied. - [0090]In STEP
**15**, weights and an inverse FFT in K_{X }are applied to compress the data in X, forming a final focused image**50**. - [0091]If SVA is used with the input spectrum of minimum length, processing steps can be reduced.
- [0092]In STEP
**16**, an inverse FFT in K_{Y }is applied to compress the data in Y without weighting so that SVA can be applied in STEP**17**. After this step, the data is in the image domain (X, Y). - [0093]In STEP 17, a two dimensional SVA is applied to reduce the sidelobe level while maintaining the mainlobe width minimum. The SVA function outputs a final focused image
**50**. - [0094]Phase correction becomes much simpler when the processed image is oriented in the range-azimuth direction. This is because there is no need to include the phase adjustment by Δθ to align the direction of smearing to the vertical. As a result, steps
**3**,**10**,**11**,**13**, and**16**in FIG. 13 can be eliminated. Since (φ_{1}−φ) is zero, ΔΦ that is expressed in Eqn. (9) is not a function of the range variable Y any more. Simplified processing steps with these adjustments from FIG. 13 are illustrated in FIG. 14. - [0095][0095]FIG. 14 is a flow chart showing the simplified processing sequence for focusing RMA-based SAR images oriented in range-azimuth. The sequence is similar to that of FIG. 13 with STEPS
**3**,**10**,**11**,**13**, and**16**eliminated. After autofocusing (STEP**7**), if the input data was of extended length, then STEPS**8**,**9**and**14**are performed. Otherwise, those steps are skipped. Then STEPS**15**and**17**are applied, generating the final focused image**50**. - [0096]Thus, the present invention has been described herein with reference to a particular embodiment for a particular application. Those having ordinary skill in the art and access to the present teachings will recognize additional modifications, applications and embodiments within the scope thereof.
- [0097]It is therefore intended by the appended claims to cover any and all such applications, modifications and embodiments within the scope of the present invention.
- [0098]Accordingly,

Referenced by

Citing Patent | Filing date | Publication date | Applicant | Title |
---|---|---|---|---|

US7236125 * | Jan 25, 2005 | Jun 26, 2007 | Saab Ab | Method for reducing angular blur in radar pictures |

US7245250 * | Aug 16, 2005 | Jul 17, 2007 | Itt Manufacturing Enterprises, Inc. | Synthetic aperture radar image compression |

US8044846 * | Nov 29, 2007 | Oct 25, 2011 | Lockheed Martin Corporation | Method for deblurring radar range-doppler images |

US9329264 | Feb 15, 2013 | May 3, 2016 | Raytheon Company | SAR image formation |

US20060066475 * | Jan 25, 2005 | Mar 30, 2006 | Saab Ab | Method for reducing angular blur in radar pictures |

CN103837862A * | Feb 20, 2014 | Jun 4, 2014 | 北京理工大学 | Method for restraining grating lobes in stepped frequency SAR distance direction based on image |

WO2014126638A1 | Dec 9, 2013 | Aug 21, 2014 | Raytheon Company | Sar image formation |

Classifications

U.S. Classification | 342/25.00R, 342/196, 342/162, 342/192 |

International Classification | G01S13/90 |

Cooperative Classification | G01S2013/9041, G01S13/9035, G01S13/9011 |

European Classification | G01S13/90S |

Legal Events

Date | Code | Event | Description |
---|---|---|---|

May 31, 2002 | AS | Assignment | Owner name: RAYTHEON COMPANY, MASSACHUSETTS Free format text: ASSIGNMENT OF ASSIGNORS INTEREST;ASSIGNOR:CHO, KWANG M.;REEL/FRAME:012965/0467 Effective date: 20020531 |

May 18, 2007 | FPAY | Fee payment | Year of fee payment: 4 |

May 11, 2011 | FPAY | Fee payment | Year of fee payment: 8 |

May 27, 2015 | FPAY | Fee payment | Year of fee payment: 12 |

Rotate